PHYSICAL REVIEW C 90, 014302 (2014)

Merging of the islands of inversion at N = 20 and N = 28

E. Caurier,1 F. Nowacki,1 and A. Poves2,3

1
IPHC, IN2P3-CNRS and Université Louis Pasteur, F-67037 Strasbourg, France
2
Departamento de Fı́sica Teórica and IFT-UAM/CSIC, Universidad Autónoma de Madrid, E-28049 Madrid, Spain
3
Isolde (CERN) 1211 Genève 23, Switzerland
(Received 1 April 2014; revised manuscript received 3 June 2014; published 10 July 2014)

The N = 20 and N = 28 “islands of inversion” are described by large scale shell model calculations with
an extension of the interaction SDPF-U that makes it possible to mix configurations with different Nω or
equivalently with different numbers of particles promoted from the sd shell to the pf shell. It allows to connect
the classical sd-shell calculations below N = 18 with the sd (protons)-pf (neutrons) calculations beyond
N = 24–26, for all the isotopes from oxygen to sulfur, using the same interaction. For some isotopes this range
contains all the nuclei between the proton and the neutron drip lines and includes the N = 20 and N = 28 islands
of inversion. We pay particular attention to the properties of the states at fixed Nω which turn out to be the real
protagonists of the physics at N = 20. The existence of islands of inversion or deformation are explained as the
result of the competition between the spherical mean field which favors the 0ω configurations and the nuclear
correlations which favor the deformed N ω configurations. The magnesium chain is exceptional because in it
the N = 20 and N = 28 islands of inversion merge, enclosing all the isotopes between N = 19 and N = 30.
Indeed, this would be also the case for the neon and sodium chains if their drip lines would reach N = 28.

DOI: 10.1103/PhysRevC.90.014302 PACS number(s): 21.60.Cs, 23.40.−s, 27.40.+z

I. INTRODUCTION The interaction SDPF-U [11] that we proposed some time

ago was aimed to describe the very neutron rich nuclei around
At the neutron rich edge, the structure of the spherical mean
N = 28 in a 0ω space, with valence protons in the sd shell
field may be at variance with the usual one at the stability line.
and valence neutrons in the pf shell. Therefore, it is applicable
The reason is that, at the stability line, the T = 0 channel of
only to nuclei with 8  Z  20 and 20  N  40 and does
the nucleon-nucleon interaction has a stronger weight relative
not describe intruder states. The main asset of SDPF-U was
to the T = 1 channel than it has when the neutron excess is
the description of the vanishing of the N = 28 shell closure
very large. If the spherical mean field gaps get reduced, open
below 48 Ca, most notably in 42 Si [12] (a result which is
shell configurations, usually two neutron excitations across the
now fully verified [13], but which produced initially some
neutron closure, take advantage of the availability of open shell
heated debates [14]). 42 Si was predicted to be oblate deformed
protons to build highly correlated states that can be more bound
and 40 Mg prolate deformed, exhibiting perhaps a neutron
than the closed shell configuration. Then the shell closure is
halo. Since its publication, it has been frequently used and
said to have vanished. Although it has long been known that
shown to give an excellent description of this region of very
the ground state parity of 11 Be was at odds with the naive
neutron rich nuclei [15]. Very recently, these calculations
shell model picture [1], this fact was overlooked until much
were repeated in the same valence space with a somewhat
later, in connection with the discovery of halo nuclei with
different effective interaction, getting (as could be expected)
N = 8. Studies of charge radii, atomic masses, and nuclear
very similar results [16]. As the sd part of SDPF-U is just
spectra in the Mg and Na isotopic chains did show that a
the USD interaction [17] and its pf part a minor variant of
region of deformation exists around N = 20 below 34 Si. Key
KB3 previous to KB3G, it is appealing to complete SDPF-U
experimental references are gathered in Refs. [2–5]. Since
with the sd-pf off-diagonal matrix elements and to retune
then, a lot a experimental and theoretical work has ensued.
the sd-pf cross shell monopoles in such a way that the
Early mean field calculations suggested that deformation was
SDPF-U results at 0ω are mostly preserved and the sd-pf
responsible for the excess of binding of 31 Na [6], but at this
gaps are in accord with the experiment. This process results
stage to get a deformed minimum required the inclusion of
in the SDPF-U-MIX interaction. More details are given in the
ad hoc rotational corrections. In the framework of the shell
Appendix. The calculations are carried out using the codes
model, the deformation in the region was soon associated with
ANTOINE and NATHAN [18] and reach basis dimensions of
the dominance of two-particle–two-hole (2p-2h) excitations
O(1010 ). In a (very) loose sense one can pretend that this
across the N = 20 shell gap between the normally occupied
interaction covers the sector of the Segré chart 8  Z,N  40.
neutron d3/2 orbit and the valence f7/2 and p3/2 orbits [7].
In this article we concentrate on the physics of the N = 20
These configurations were dubbed intruders since they do not
“island of inversion” and its merging in some cases with the
obey the normal filling of the standard spherical mean field.
neighboring N = 28 one.
More recent shell model works include the Monte Carlo Shell
Model (MCSM) calculations of the Tokyo group [8] and other
large scale calculations in the sd-pf valence space [9]. Beyond II. THE PHYSICS AT FIXED Nω
mean field calculations have also been used in the description What is the driving force behind the abrupt changes leading
of the region with diverse degrees of success [10]. to the appearance of these islands of inversion? What makes

0556-2813/2014/90(1)/014302(9) 014302-1 ©2014 American Physical Society

E. CAURIER, F. NOWACKI, AND A. POVES
these intruder states special is that they need to be highly
correlated in order to compensate for the energy loss associated
to the breaking of the normal filling of the spherical mean
field. Obviously, small gaps are easier to overcome; thus,
a reduction of the neutron magic gaps at the very neutron
rich edge is good news for the intruders. The mechanisms
need not be the same in the different regions. For instance,
in 11 Li the intruder is mostly pairing boosted while in 11 Be
the quadrupole interaction is more important. In the other
three neutron rich regions, N = 20, N = 28, and N = 40, the
quadrupole interaction is the main player. Let us concentrate
on the N = 20 case. Compared to the configurations with
closed N = 20, the intruders (np-nh) have neutrons in open
sd- and pf -shell orbits and in some cases protons in open
sd-shell orbits. This favors the efficient buildup of correlations
by the neutron-proton quadrupole interaction when the open
orbits are the appropriate ones. And whose are these is
dictated by the different variants of SU(3). For instance,
when valence neutrons or protons occupy quasidegenerate
orbits with jr − js = 2 and lr − ls = 2 the coupling scheme
is quasi-SU(3) [19], if they are in quasispin doublets the
regime is that of pseudo-SU(3) [20]. In the limit of vanishing
spin-orbit splitting, all the orbits in a harmonic oscillator shell
form Elliott’s SU(3) multiplet [21]. To get large coherence the
neutrons and the protons must pertain to one or another of these
coupling schemes. For example, in the case of the N = 20
intruders, the neutrons in the orbits 0f7/2 and 1p3/2 , and the
protons in 0d5/2 and 1s1/2 , are in the quasi-SU(3) regime and
the neutrons in 0d3/2 and 1s1/2 are in pseudo-SU(3).
Let us make these statement quantitative in a few selected
cases. In this section all the calculations are performed at fixed
N ω. We only allow neutron jumps from the sd to the pf
shell without any other truncation. We have verified that the
effect of the proton excitations to the pf shell is negligible
below Z = 16. We take care of the (small) center-of-mass
contamination by adding to the effective interaction the center-
of-mass Hamiltonian (with ω = A). The expectation value of
the center-of-mass Hamiltonian in the physical states is always
below 0.001A. The results for the low energy levels of 32 Mg
are presented in Fig. 1. We can follow the evolution from the
semimagic 0p-0h result, with a high excited 2+ and a low
4 4+
Excitation energy (MeV)
16
3 quadrupole moment. If we go to the SU(3) limit in the pf -shell
2+
4+
2 36
+ 107
0
1 2+ 168
83
+
0 121
0 0
0p-0h 2p-2h 4p-4h
FIG. 1. (Color online) The low energy spectra and B(E2)’s
(in e2 fm4 ) of the 0p-0h, 2p-2h, and 4p-4h configurations in 32 Mg.
PHYSICAL REVIEW C 90, 014302 (2014)
B(E2) to a rotational-like 2p-2h whose B(E2) corresponds
to β = 0.4/0.5 and finally to a perfect rigid rotor 4p-4h with
E(4+ )/E(2+ ) = 3.2 and a very large B(E2) that corresponds
to a superdeformed structure. Most important for our aims
is that the gains in energy due to the correlations—defined
as the difference between the energy which comes out of
the diagonalization and the energy of the lowest 0+ state of
seniority zero in the corresponding space—are very different
in the 0p-0h, 2p-2h, and 4p-4h spaces: 1.5, 12.5, and 21 MeV,
respectively. These huge correlation energies may eventually
overcome the spherical mean field gaps. In fact this is the
case in 32 Mg. With SDPF-U-MIX the lowest 4p-4h 0+ state
is about 250 keV below the lowest 0+ of the 2p-2h space
and 1.2 MeV below the 0+ of the 0p-0h configuration. This
near degeneracy of the 2p-2h and 4p-4h bandheads is not
a spurious manifestation of our spherical mean field not
producing the right sd-pf gap; rather it is due to the fact
that the energy gain per particle promoted to the pf shell is
the same for both configurations. We want to stress again the
fact that, in favorable circumstances like these, the gain in
correlation energy of the intruders can beat the spherical mean
field. In fact, in the laboratory frame, this is the microscopic
mechanism responsible for the shape transitions from spherical
to deformed nuclei [19]. The lowest negative parity state of
1p-1h nature is a 3− , 4 MeV above the 2p-2h 0+ , and the
lowest 3p-3h state is a 2− , 2.5 MeV above the 2p-2h 0+ .
Their respective energy gains are 5 and 16 MeV and the
underlying structures correspond to the K = 3− and K = 2−
bandheads as expected from the Nilsson diagrams for β = 0.15
and β = 0.4.
The 4p-4h state of 32 Mg has an academic interest in itself
even if the states belonging to its rotational band do not
manifest themselves openly in the low energy spectrum (as
do their cousins in the superdeformed bands of 36 Ar and
40
Ca [22–24]) because of its strong mixing with the 0p-0h and
2p-2h spherical and deformed states. It may well happen that
they could become yrast at some higher spin, but the threshold
for neutron emission is not very high, and the experiments to
find them are probably hopeless. In fact, one can understand
semiquantitatively why this configuration can produce such
superdeformed structure in the context of Elliott’s SU(3) and
its variants. Let us assume that the four pf -shell neutrons
are in the quasi-SU(3) scheme and the four neutron holes in
sd are in the pseudo-SU(3) scheme; in this case the neutrons
contribute with 24b2 (times the effective charge) (where b
is the harmonic oscillator length parameter) to the intrinsic
sector this number increases to 26b2 . The value from the shell
model calculation is 24.7b2 . For the protons, the quasi-SU(3)
limit gives 11b2 against 9.7b2 (times the effective charge) of
4+ the shell model calculation. With effective charges 0.46 and
1.31 for neutrons and protons, taken from the work of Dufour
2+
and Zuker [25], these values lead to β = 0.6/0.7 depending
+
of the definition of β.
It follows from the above discussion that the configuration
with four neutrons in the pf shell and two neutron holes in the
sd shell maximizes the quadrupole moment and, a fortiori, the
quadrupole correlation energy. Therefore, one should expect
the 2p-2h configurations to be also dominant in 34 Mg. On the
014302-2
MERGING OF THE ISLANDS OF INVERSION AT N = 20 . . . PHYSICAL REVIEW C 90, 014302 (2014)

_
1.4 1/2
1.2

Excitation energy (MeV)

0.8
Excitation energy (MeV)

+
5/2
1
_
0.6 0.8 7/2 +
_
7/2
7/2 + +
3/2 0.6
_ 5/2
_ 5/2
0.4 3/2
0.4

0.2 _
0.2 3/2 3/2
+
+ 0 +
3/2 1/2
+ 2p-2h 3p-3h
0 1/2
2p-2h 1p-1h 0p-0h FIG. 3. (Color online) The low energy spectra of the 2p-2h and
+
3p-3h configurations in 33 Mg. Energies are relative to the 3p-3h 12
FIG. 2. (Color online) The low energy spectra of the 0p-0h,
state.
1p-1h, and 2p-2h configurations in 31 Mg. Energies are relative to
+
the 2p-2h 12 state.
+
of 100 e2 fm4 . In particular the K = 12 3p-3h band can be
contrary, one expects the 0p-0h ones to begin taking over in viewed as the addition of two neutrons to the ground state
36
Mg. This would establish the limit of the N = 20 island of band of 31 Mg. It turns out that both bandheads, in spite of their
inversion. However, as we see in the next section, the very large different spin and parity, have negative magnetic moments
depopulation of the 0f7/2 orbit in 36 Mg indicates that before (−0.49μN for the 2p-2h and −0.87 for the 3p-3h). Contrary
leaving the N = 20 island of inversion we enter another, the to the assumption of Ref. [27], the magnetic moment of the
3+
N = 28 one, meaning that both islands are actually merged 2
3p-3h state is positive (+0.62μN ). The results of the fully
in a single one. These arguments apply as well to the 3p-3h mixed calculation, which favor negative parity for the ground
excitations in the N = 21 isotopes, which we expect to be very state, are discussed in Sec. VI.
low in energy. Similar analysis can be carried out for all the remaining
In 31 Mg the configurations 0p-0h, 1p-1h, and 2p-2h are isotopes. We want to underline here two important points:
nearly degenerate. The lowest one is the 2p-2h configuration, (i) The configurations at fixed np-nh contain much of the
+ +
which looks like a K = 12 band, with an excited 32 at relevant physics, and (ii) when configurations with different
∼100 keV, in agreement with the experimental findings of particle-hole structures, and hence with very different amounts
Refs. [4]. The energy gain of the band is 14.5 MeV. The lowest of energy gains due to the correlations, compete, as is the
+
0p-0h state, a 32 , gains just 3.5 MeV and is 400 keV less case for the states of different parities in some N = 19 and
+ N = 21 isotopes, the final balance between monopole energy
bound than the 2p-2h 12 . The lowest 1p-1h negative parity losses and correlation gains is very delicate and the difficulty
− +
state, a 32 , gains 8.5 MeV and is 400 keV above the 2p-2h 12 . in accounting for experimental energy splittings between the
These results are gathered in Fig. 2. The E2 and M1 transition positive and negative parity bands which may be smaller than
probabilities of the 2p-2h band compare well with the recent 100 keV is extreme.
experimental values from Ref. [26]:
5+ 3+
  
B(M1) 2
→ 2
= 0.1–0.5μ2N theor. 0.35μ2N , III. SPHERICAL MEAN FIELD VERSUS CORRELATION
3+ 1+
   ENERGIES: THE MECHANISM
B(M1) 2 → 2
= 0.019(4)μ2N theor. 0.03μ2N , OF CONFIGURATION INVERSION
 + 1+

B(E2) 52 → 2
= 61(7) e2 fm4 (theor. 84e2 fm4 ). As we have already anticipated, the islands of inversion
occur when a group of adjacent nuclei have their ground
+ states dominated by intruder configurations. We develop now
The magnetic moment of the 12 (using bare g fac-
tors) is −0.85μN , very close to the experimental value the case of the N = 20 isotopes. We plotted in Fig. 4 the
−0.88355(15)μN ; thus, we can expect that its 2p-2h character correlation energies of the lowest states of the 0p-0h and
is rather pure, the more so in view of the absence of nearby 2p-2h configurations. As the uncorrelated energy we take in
1+ each case the lowest diagonal energy (expectation value of
2
states to mix with. The intrinsic quadrupole moment of the
the Hamiltonian) in a basis of states coupled to good J and
ground state band is typical of this region, Q0 ≈ 70 e fm2 . with well defined generalized seniority. Because of this choice
In 33 Mg the lowest state at fixed configuration is the 3p-3h we incorporate in fact some diagonal pairing energy in our
1+ +
2
, head of a K = 12 band. At 150 keV appears the 2p-2h uncorrelated reference, but this is irrelevant for our purpose. As
3− −
2
, head of a K = 32 band. The 0p-0h and 1p-1h states lie expected for semimagic nuclei, for the 0p-0h configurations,
more than 1.5 MeV higher. These results are gathered in Fig. 3. the correlation energies are small and roughly constant. On
Both structures are highly collective, with B(E2)’s in excess the contrary, for the 2p-2h intruders, they can be very large

014302-3
E. CAURIER, F. NOWACKI, AND A. POVES PHYSICAL REVIEW C 90, 014302 (2014)

0 2
EXP
TH

2 excitation energy (MeV)

1.8
Correlation energy (MeV)

-2

-4 1.6
0p-0h
1.4
-6 2p-2h
1.2
-8
1
-10

-12
-14
8 9 10 11 12 13 14 15
Z N
FIG. 4. (Color online) Correlation energies of the
(squares) and 2p-2h (circles) configurations at N = 20.
and have a rapid variation with Z. The largest values occur at
mid proton shell, when the quadrupole collectivity reaches its
maximum.
In Fig. 5 we present the differences in energy between the
lowest 0p-0h state with well defined J π and the lowest 2p-2h
state without correlations. It is seen that in all cases the normal
filling gives the lowest energy, although between Z = 8 and
Z = 14 there is an almost linear increase from 3 to 12 MeV,
while from there on the curve is much flatter. This reflects
the reduction of the sd-pf gap as we approach the neutron
drip line. When we take fully into account the correlations the
situation changes dramatically as reflected in the lower curve
of the figure. The balance between the correlation gains and the
monopole losses of energy defines the borders of the island of
inversion at N = 20 in 29 F and 33 Al. Clearly, 30 Ne, 31 Na, and
32
Mg are bona fide members of the club. Equivalent graphs can
be drawn for the other isotonic chains. Roughly speaking the
E(2p-2h)-E(0p-0h) : without correlations
E(2p-2h)-E(0p-0h) : with correlations
14
12
10
Gap (MeV)
+
0.8
0.6
12 14 16 18 20 22 24 26 28 30 32
16 17 18
0p-0h FIG. 6. (Color online) Excitation energies of the first 2+ state in
the magnesium isotopes. Results of the calculations with the SDPF-
U-MIX interaction in the valence space of the sd shell for the protons
and the sd-pf shells for the neutrons, compared with the available
experimental data.
situation is very similar for the N = 19, N = 21, and N = 22
isotonic chains. It is probably not worthwhile to go much
more beyond this qualitative definition of the somewhat fuzzy
shores of the island of inversion around N = 20 because the
predictions obtained in the analysis at fixed configuration may
sometimes change when the full mixing is taken into account,
the more so for the nuclei near to the borders. We are more
precise in the section dealing with the full scale results of our
calculations.
IV. FROM N = Z TO N = 32 IN THE Mg, Ne,
AND Si ISOTOPES
In Fig. 6 we compare the experimental 2+ excitation
energies of the even Mg isotopes, starting at N = Z, with the
shell model calculations with the SDPF-U-MIX interaction.
Up to N = 16 the results should not differ much from the
ones produced by the USD interactions [17]. Beyond N = 16
the calculations include (if necessary for convergence) up
to 6p-6h excitations from the sd shell to the full pf . The
agreement is excellent and covers the span of isotopes from
24
Mg to the neutron drip line. Notice the disappearance of the
semimagic closures at N = 20 and N = 28 and the presence

8
of a large region of deformation which connects the two
6 islands of inversion, previously thought to be split apart. The
4 agreement is really superb. Beyond N = 24 the effect of the
core excitations is perturbative and produces a small expansion
2 ISLAND OF INVERSION of the spectra which improves slightly the agreement with
0 the experimental data obtained in the 0ω calculations. The
merging of the N = 20 and N = 28 islands of inversion is
-2 evident.
8 9 10 11 12 13 14 15 16 17 18
Z In Fig. 7 we compare the B(E2)’s in the transition region
with the experimental data including some unpublished results
FIG. 5. (Color online) The gap between the 0p-0h and the from RIKEN [28]. We use effective charges of 1.35 and
2p-2h configurations at N = 20, without correlations (squares) and 0.35 for protons and neutrons, respectively, which are fully
including correlations (circles). Nuclei close to or below the zero line compatible with a recent fit to the sd-shell nuclei with the
are candidates to belong to the island of inversion. interaction USD-A [29] and with the results obtained by

014302-4
MERGING OF THE ISLANDS OF INVERSION AT N = 20 . . . PHYSICAL REVIEW C 90, 014302 (2014)

150 2.5

exp
th

2 excitation energy in MeV

2

100
B(E2) (e fm )
4
2

1.5

50 B(E2) exp
B(E2) th 1

+
0 0.5
30 32 34 36 38 40 42 8 10 12 14 16 18 20 22 24 26 28 30 32
A N

FIG. 7. (Color online) B(E2)’s of the magnesium isotopes com- FIG. 9. (Color online) Excitation energies of the first 2+ states in
pared to the experimental results. the neon isotopes (see caption of Fig. 6).

Dufour and Zuker in Ref. [25]. We take ω = 45A−1/3 −
25A−2/3 . The agreement is very good as well.
In Fig. 8 we gathered the occupancies of the pf -shell
orbits in the even-even magnesium isotopes. We aggregated
the values of the f and p orbits for simplicity. The reference
numbers for the total pf -shell occupancies are those labeled
0ω in the figure. The pf shell has more than two neutrons
in excess at N = 20 and N = 22. At N = 24 the excess is of
about one neutron, and beyond that, the core excitations are
much damped.
What is more interesting is that when the sd-shell core
excitations become small, the occupancy of the p orbits
(mainly 1p3/2 ) keeps increasing so that in N = 26 and N =
28 about two neutrons are in 1p3/2 , whereas the expected
occupancy if N = 28 were a strong closure would have been
zero. In this sense we can speak of the merging of the islands of
inversion at N = 20 and N = 28 in the magnesium isotopes.
Notice also that a large occupancy of the p orbits favors the
appearance of a neutron halo when the neutron separation
energy becomes close to zero, as might be the case in 37,39 Mg
and 40 Mg. Our occupancies for the pf -shell orbits in 32 Mg
agree with the experimental results of Ref. [30].
The results for the neon isotopes (Fig. 9) are very similar to
that for magnesium, although in this case the N = 28 isotope
38
Ne is most probably beyond the neutron drip line. The 2+
excitation energy of 32 Ne is taken from Ref. [31]. In Fig. 10 we
collected the occupancies of the f and p orbits in the isotopic
chain as a function of the neutron numbers. The behavior is
very similar to that in the magnesium chain, except that the
p orbits are even more occupied. We have added the numbers
for 31 Ne, because some recent experimental data [32] suggest
that it could develop a neutron halo. Indeed, our results are
consistent with this hypothesis because the 1p3/2 orbit has on
average more than one neutron.
In Fig. 11 we show the results for the silicon isotopes
(notice the very different energy scale). At variance with the
magnesium case, we observe a majestic peak at N = 20, a

10 8 nf
nf
np
np
npf
Number of neutrons

8 npf
Number of neutrons

6 0hw
0hω
6
4
4
2
2

0 0
30 32 34 36 38 40 42 28 30 32 34 36 38
A A
FIG. 8. (Color online) Occupation numbers of the pf -shell orbits FIG. 10. (Color online) Occupation numbers of the pf -shell
in the magnesium isotopes; f orbits (black), p orbits (red), pf shell orbits in the neon isotopes: f orbits (black), p orbits (red), pf shell
(blue), and normal filling (green). (blue), and normal filling (green).

014302-5
E. CAURIER, F. NOWACKI, AND A. POVES
4 4
2 excitation energy (MeV)
3 3
2 0
1
+
0
14 16 18 20 22 24
N exp Mg th
FIG. 11. (Color online) Excitation energies of the first 2+ states
in the silicon isotopes (see caption of Fig. 6).
fingerprint of the double magic nature of 34 Si which we discuss
in more detail later, and, as in the neon and magnesium cases,
no trace of the N = 28 shell closure is seen, in agreement
with the findings of recent experiments at GANIL [12] and
RIKEN [13].
Some of these results were published in the proceedings
of Ref. [33] but were mistakenly attributed to the SDPF-U-SI
interaction; thus, we offer this erratum.
V. LANDING AT THE ISLAND OF INVERSION:
30
Mg →32 Mg AND 34 Si →32 Mg
There are two courses to land at the island of inversion by
the 32 Mg shore: through the isotopic and the isotonic chains.
Both are of paramount importance for the understanding of
the rich variety of structural changes which take place in the
region. Adding two neutrons to 30 Mg provokes the inversion
of the normal and intruder configurations which are shifted by
nearly 3 MeV in 32 Mg. In the isotonic course the transition
is even more abrupt, as was recently shown in a GANIL
experiment [34]: by removing two protons from 34 Si, the
intruder (deformed) state is shifted down by about 4 MeV
with respect to the spherical one to become the ground state
of 32 Mg.
We compare the experimental data with the shell model
results in Fig. 12. The calculations include configurations with
up to six neutrons in the pf shell. 30 Mg and 34 Si have ground
states which are dominantly (>80%) 0p-0h and first excited
0+ ’s dominantly 2p-2h. They differ in the structure of the
lowest 2+ which is 0p-0h in 30 Mg and 2p-2h in 34 Si. More
details on this last nucleus can be found in Ref. [34], where a
close to final version of SDPF-U-MIX was utilized.
The structure of the 0+ states in 32 Mg is extremely singular;
the ground state has 9% 0p-0h, 54% 2p-2h, 35% 4p-4h, and
1% 6p-6h; thus, it is a mixture of deformed and superdeformed
shapes. The excited 0+ state has 33% 0p-0h, 12% 2p-2h, 54%
4p-4h, and 1% 6p-6h, a surprising hybrid of spherical and
superdeformed shapes, whose direct mixing matrix element is
PHYSICAL REVIEW C 90, 014302 (2014)
EXP + +
4 2
TH
Excitation energies (MeV)
+
+
0
4
2 +
+
2 +
0
1 +
2
+ + +
0 0 0 0
26 28 30 32 30 32 34
exp Mg th exp Si th
FIG. 12. (Color online) Comparison between experiment and
theory for the most important low lying states in 30 Mg, 32 Mg, and
34
Si.
strictly zero. One could fancy to name it the shape entangled
state.
The 2+ state has a structure similar to the ground state. As
shown in Fig. 7 its B(E2) agrees with the experimental result.
In addition, the calculated spectroscopic quadrupole moments
of the 2+ and 4+ states and the B(E2)’s in the yrast band are
compatible with a single intrinsic state with Q0 ≈ 65 e fm2 .
The MCSM calculations of Ref. [8], which only include the
0f7/2 and 1p3/2 orbits of the pf shell, give results similar to
ours except for the excited 0+ state, which is too high by almost
2 MeV. Similarly, in 34−40 Mg the calculated E2 properties are
compatible with Q0 ≈ 70 e fm2 , which is another fingerprint
of the merging of the N = 20 and N = 28 islands of inversion
or deformation.
Since the early β-decay experiments at Isolde [35] it
is known that in 32 Mg there are many states, mostly of
negative parity, above the 4+ state. They have been explored
more recently via the 32 Na β decay [27,36] or in (p,p  )
experiments [37]. Reference [27] presents also the MCSM
predictions for the negative parity states fed in the β decay.
The experimental level at 2.551 MeV is most probably the
second 2+ state. MCSM puts it at 3 MeV, whereas we get
it at nearly the same energy as the 4+ state. According to
these references, the lowest experimental negative parity state
would appear at 2.858 MeV. The calculated negative parity
states are 1− at 3.0 MeV, 2− at 3.1 MeV, 3− at 3.4 MeV, 4− at
3.9 MeV, 0− at 4.0 MeV, and 5− at 4.2 MeV. They are mostly of
3p-3h nature. The lowest negative parity states in the MCSM
description are of 3p-3h nature as well, and start at 3.8 MeV
with four close packed states (2− , 1− , 2− , 3− ) followed by a
doublet (4− , 5− ) at about 4.5 MeV.
VI. MISCELLANEOUS RESULTS
31
A. Mg and 33 Mg
The N = 19 and N = 21 isotonic chains are very complex,
because of the near degeneracy of configurations with different
particle-hole structure, as discussed in Sec. II. In 31 Mg the
014302-6
MERGING OF THE ISLANDS OF INVERSION AT N = 20 . . . PHYSICAL REVIEW C 90, 014302 (2014)

TABLE I. Excitation energies (in MeV) and magnetic moments 1.5

(in μN ) for the low lying states of 31 Mg and 33 Mg with (b)are and
(e)ffective g factors. +

Excitation energies (MeV)

7/2 + +
7/2 7/2 7/2
+
π
J E (expt.) E (theor.) μ (theor.) (b) μ (theor.) (e) 1
31 1+
Mg 2
0.0 0.04 –0.93 –0.65
3+
2
0.05 0.04 +1.13 +0.81 + +
3− 0.5 5/2 , 3/2 +
2
0.221 0.0 –1.24 –1.07 + + 5/2
5/2 5/2
33 3−
Mg 2
0.0 0.0 –0.54 –0.49
5− + + +
2
0.484 0.33 –0.07 –0.09 + + 3/2 , 5/2 3/2
0 3/2 3/2
1+
2
0.04 –0.93 –0.69
3+
2
0.12 +0.69 +0.44 31 31 33 33
Na (exp) Na (th) Na (exp) Na (th)

FIG. 13. (Color online) Comparison of the calculated spectra

situation is especially critical. We have not tried to fine-tune
the interaction to improve our results which amount to having
+ + −
the 12 , 32 , and 32 states degenerated (see Table I). The
+ C.
magnetic moment of the 12 ground state was measured as
−0.88355(15)μN [4]. Our predictions show a very strong
dependence on the choice of the gyromagnetic factors and
marginally agree with the experimental value.
− 2.235 MeV, respectively. The B(E2), (2+ →0+ ) is predicted at
In 33 Mg the fully mixed calculation produces a 32 ground
+
state with the 12 state just 40 keV higher. The magnetic
−
moment of the 32 , −0.54μN (b), −0.49μN (e), is short
from the experimental value, −0.7456(5)μN . If the J =
3/2 experimental assignment is firm, then positive parity
is excluded by the sign of the magnetic moment. Notice,
however, that our results locate the positive parity states almost
degenerated with the ground state. The magnetic moment of
−
the 4p-4h 32 state is −1.67μN (bare) [−1.36μN (effective)];
therefore, a somewhat larger mixing of 4p-4h components than
that given by our calculation may line up the theoretical value
with the experimental one.
with the available experimental data in 31 Na and 33 Na.
30
Ne, 31 Ne, and 32 Ne
To complete this spectroscopic tour, we visit the neon chain.
Our results for 30 Ne, 2+ at 0.79 MeV and 4+ at 2.14 MeV,
compare very well with the experimental values, 0.792 and
66 e2 fm4 compared with the measured value 90(54) e2 fm4 .
Again a single intrinsic state with Q0 ∼ 60 e fm2 explains
the calculated E2 properties. 32 Ne is more deformed: Q0 ∼
65 e fm2 . The calculated 2+ excitation energy, 0.67 MeV,
fits well with the experimental value, 0.72 MeV. The 4+
is predicted at 1.89 MeV. Finally, for 31 Ne the calculation
−
produces a 32 ground state which is mainly 2p-2h, and the
+
first excited state at ∼200 keV is a 32 of 3p-3h character,
+
belonging to the K = 12 in parallel with what happens in its
−
isotone 33 Mg. The intrinsic quadrupole moment of the K = 32
band is Q0 ∼ 60 e fm , and, as we mentioned before, the
2

occupation of the p orbits exceeds 1.2 neutrons.

31
B. Na and 33 Na
29
D. F and 31 F
Na was for many years the protagonist of the N = 20
31

saga, even if only the properties of its ground state were known
(spin parity, magnetic moment, isotope shift, binding energy).
Although other nuclei have taken up the relay nowadays,
it still deserves attention. We have gathered the available
experimental information in Fig. 13. The newest data [38,39]
consist of the excitation energies of two members of the K =
3+
2
ground state rotational band and the B(E2) of the lowest
in-band decay in 31 Na, and the excitation energies of two levels
in 33 Na. Notice the very nice agreement of the calculation and
the data which extends to the ground state magnetic moment
[2.298μN (experiment) vs the calculated 2.26μN (b) or 1.96μN
(e)]. As in the 32 Mg case the calculated E2 properties of 31 Na
are compatible with an intrinsic state with Q0 ∼ 65 e fm2 . We
have also plotted the results for 33 Na in which the behavior of
the calculated excitation energies is closer to J (J + 1) than in
the previous case, and Q0 ∼ 72 e fm2 . The comparison of the
new data with the calculated values is quite good and supports
+
strongly a 32 ground state.
Not very much is known experimentally about 29 F and 31 F.
+
Our calculations produce a 52 ground state in 29 F, which is
+
60% 0ω with a first excited 12 at 0.91 MeV, which is 80%
+
intruder and the head of a K = 12 band, as expected from
quasi-SU(3) and Nilsson diagrams. This compares fairly well
with a recent measure at RIKEN [40], which places this state
at 1.06 MeV. In our calculation, the ground state of 31 F is an
+ +
extremely mixed 52 (66% intruder) and the excited 12 (74%
intruder) appears at much lower excitation energy, 0.21 MeV.
Neutron excitations result in binding energy gains of 1.9 and
2.5 MeV, respectively, which may help to explain the far off
location of the fluorine neutron drip line.
33
E. Al and 35 Al
33
Al has its ground states largely dominated by the “normal”
configurations (∼80%). Thus, it does not belong properly
to the island of inversion. The calculations reproduce very

014302-7
E. CAURIER, F. NOWACKI, AND A. POVES
+ the study of the one and two neutron separation energies,
well the properties of the 52 ground state; the magnetic
moment is +4.088(5)μN vs +4.17μN (b) and +3.86μN (e);
the spectroscopic quadrupole moment is +0.12 eb compared
to the calculated +0.12 eb. Our results do not produce a low
+
lying 52 as surmised in the experiment of Ref. [41]. This
is consistent with the large excitation energy of the intruder
+
0+ in 34 Si. The lowest excited state is predicted to be a 52
of intruder nature at 1.70 MeV followed by two other 2p-2h
+ +
states at 1.85 MeV ( 12 ) and 2.28 MeV( 52 ). Contrary to some
compiled results we do not produce negative parity states in
+ S2009-ESP-1473). A.P. is supported by MINECOs (Spain)
this range of energies. In the 52 ground state of 35 Al the
“normal” configurations still lead, but barely so at 52%. The
+ +
lowest excited states, 12 at 0.63 MeV and 32 at 0.80 MeV,
are intruders.
F. The limits of the “big island of deformation”
We argued already that the previously established N = 20
and N = 28 islands of inversion or deformation merge in
the neon, sodium, and magnesium isotopic chains, creating a
bigger one (big island of deformation, or BID). Referring only
to the ground states, their N = 19 isotopes seem to belong
to it as well, and their N = 18 ones not (but see below).
The value N = 31 and the neutron drip line define the west
shore of the BID. Some heavy aluminums, silicons (N  26),
phosphors, and sulfurs (N  28) do belong to the N = 28
sector of the BID as well, but their less neutron rich isotopes
do not belong to the N = 20 sector (except perhaps 34 Al).
Fluorines are transitional, as we have just discussed. Of the
N = 18 isotones, only 28 Ne and 29 Na can pretend to pertain to
it with 50% and 40% of intruder components in their ground
states. For the other isotopes, although the ground states are
“normal,” quite often intruder states show up at low excitation
energy.
VII. CONCLUSIONS
We showed that the model space comprising the sd shell
for the protons and the sd-pf shell for the neutrons, together
with the effective interaction SDPF-U-MIX, make it possible
to describe a very large region of nuclei, in particular the
very neutron rich nuclei at or around the N = 20 “islands
of inversion.” In many cases the inversion of configurations
produces deformed ground state bands, hence we use the term
“islands of deformation” as well. According to our calculations
(and also to the meager experimental data available), the two
islands merge in the magnesium chain. In the calculations
they also merge in the neon and sodium chains. However, this
could only be checked experimentally if their neutron drip
lines happened to be close enough to N = 28. We studied in
detail the mechanisms that lead to the inversion of normal
and intruder configurations, paying particular attention to the
properties of the states at fixed np-nh configurations and to
their correlation energy gains. We compared the calculations to
some selected experimental results. The ubiquitous deformed
bands have intrinsic (electric) quadrupole moments in the
range Q0 = 60–80 e fm2 . We leave for the future a full scan
of the region with the SDPF-U-MIX interaction, as well as
014302-8
PHYSICAL REVIEW C 90, 014302 (2014)
which requires some extra monopole work, probably including
three-body terms.
ACKNOWLEDGMENTS
This work was partially supported by the MICINN (Spain)
(Grant No. FPA2011-29854), by the IN2P3 (France) and
MICINN (Spain) (Grant No. AIC11-D-648), and by the
Comunidad de Madrid (Spain) (Grant No. HEPHACOS
Centro de Excelencia Severo Ochoa Programme under Grant
No. SEV-2012-0249.
APPENDIX: THE SDPF-U-MIX INTERACTION
The SDPF-U-SI interaction was designed for 0ω calcu-
lations of very neutron rich sd nuclei around N = 28 in a
valence space comprising the full sd (pf ) shell for the protons
(neutrons); i.e., this interaction was defined (implicitly) with a
core of 28 O. Its single particle energies (SPEs) and monopoles
(neutron-proton sd-pf and neutron-neutron pf -pf ) were
fixed by the spectra of 35 Si, 41 Ca, 47 K, and 49 Ca. To allow
for the mixing among different np-nh neutron configurations
across N = 20, it is necessary to add to SDPF-U-SI the
following new ingredients: (a) We take the off-diagonal cross
shell sd-pf matrix elements from the Lee-Kahana-Scott G
matrix [42] scaled as in Ref. [24]. (b) We take the SPEs on
a core of 16 O: for the the sd-shell orbits we use always the
USD values [17], while for the pf -shell orbits we have no
experimental guidance at all. Nonetheless, for any particular
set of pf -shell SPEs, the neutron-neutron sd-pf monopoles
must be chosen so as to reproduce the spectrum of 35 Si and the
N = 20 gap. Because the solution is not unique, we anchored
our choice to obtain a reasonable energy for the first excited
0+ state in 30 Mg. This guarantees that in our isotopic course
toward N = 20 the descent of the intruder states proceeds with
the correct slope. (c) We incorporate into the isovector pairing
of the sd shell the same modifications introduced in Ref. [8].
The pairing reduction is needed to avoid double counting when
core excitations are taken explicitly into account. A recent
study of the effects of the change of the reference valence
space in the calculation of the renormalizations of the sd
and pf matrix elements, made in Ref. [43], gives a robust
foundation to these modifications.
A concern may arise about the applicability of this valence
space and the interaction SDPF-U-MIX to nuclei with Z >
16. Indeed, without treating properly the proton excitations,
a description of the argon and calcium isotopes close to N =
20 is excluded. Therefore, it would be premature to give an
answer before they are incorporated in the model. We expect
the neutron core excitations to do part of the work in the sulfur
isotopes near N = 20, but their effects should be somehow
blocked toward N = 28.
The very large span in neutron number that this inter-
action has to cope with brings in some global monopole
problems which can be solved by adding to the Hamiltonian
(minute) two- plus three-body monopole terms of the form
MERGING OF THE ISLANDS OF INVERSION AT N = 20 . . .
1
2
N(N − 1)V2 or 16 N (N − 1)(N − 2)V3 with N = A − 16,
which do not affect the spectroscopic results. We have not
[1] I. Talmi and I. Unna, Phys. Rev. Lett. 4, 469 (1960).
[2] C. Thibault et al., Phys. Rev. C 12, 644 (1975); G. Huber
et al., ibid. 18, 2342 (1978); C. Détraz et al., ibid. 19, 164 (1979);
D. Guillemaud-Mueller et al., Nucl. Phys. A 426, 37 (1984);
P. Baumann et al., Phys. Lett. B 228, 458 (1989).
[3] T. Motobayashi et al., Phys. Lett. B 346, 9 (1995); H. Iwasacki
et al., ibid. 522, 227 (2001); B. V. Pritychenko et al., Phys. Rev.
C 63, 011305(R) (2000); Y. Yanagisawa et al., Phys. Lett. B 566,
84 (2003); J. A. Church et al., Phys. Rev. C 72, 054320 (2005).
[4] G. Neyens et al., Phys. Rev. Lett. 94, 022501 (2005); D. T.
Yordanov et al., ibid. 99, 212501 (2007); P. Himpe et al., Phys.
Lett. B 658, 203 (2008); R. Kanungo et al., ibid. 685, 253 (2010);
G. Neyens, Phys. Rev. C 84, 064310 (2011).
[5] H. Mach et al., Phys. Lett. B 230, 21 (1989); R. W. Ibbotson
et al., Phys. Rev. Lett. 80, 2081 (1998); Ch. Hofmann et al.,
Phys. Rev. C 42, 2632 (1990); S. Nummela et al., ibid. 64,
054313 (2001); N. Iwasa et al., ibid. 67, 064315 (2003); A.
Gade et al., Phys. Rev. Lett. 99, 072502 (2007); K. Wimmer et
al., 105, 252501 (2010); N. Hinohara et al., Phys. Rev. C 84,
061302 (2011).
[6] X. Campi et al., Nucl. Phys. A 251, 193 (1975).
[7] A. Poves and J. Retamosa, Phys. Lett. B 184, 311 (1987);
E. K. Warburton, J. A. Becker, and B. A. Brown, Phys. Rev.
C 41, 1147 (1990); K. Heyde and J. L. Wood, J. Phys. G 17, 135
(1991); N. Fukunishi, T. Otsuka, and T. Sebe, Phys. Lett. B 296,
279 (1992).
[8] Y. Utsuno et al., Phys. Rev. C 60, 054315 (1999); T. Otsuka
et al., Prog. Part. Nucl. Phys. 47, 319 (2001).
[9] E. Caurier et al., Phys. Rev. C 58, 2033 (1998); J. L. Wood
et al., Nucl. Phys. A 651, 323 (1999); E. Caurier et al., ibid. 693,
374 (2001); B. A. Brown, Prog. Part. Nucl. Phys. 47, 517 (2001);
E. Caurier et al., Eur. Phys. J. A 15, 145 (2002).
[10] J. Terasaki, H. Flocard, P.-H. Heenen, and P. Bonche, Nucl. Phys.
A 621, 706 (1997); P. G. Reinhard, D. J. Dean, W. Nazarevicz,
J. Dobaczewski, J. A. Maruhn, and M. R. Strayer, Phys. Rev.
C 60, 014316 (1999); G. A. Lalazissis, D. Vretenar, P. Ring,
M. Stoisov, and L. Robledo, Nucl. Phys. A 60, 014310 (1999);
S. Peru et al., Eur. Phys. J. A 9, 35 (2000); R. Rodriguez-
Guzman, J. L. Egido, and L. M. Robledo, Phys. Rev. C 65,
024304 (2002).
[11] F. Nowacki and A. Poves, Phys. Rev. C 79, 014310 (2009).
[12] B. Bastin et al., Phys. Rev. Lett. 99, 022503 (2007).
[13] S. Takeuchi et al., Phys. Rev. Lett. 109, 182501 (2012).
[14] J. Fridmann et al., Nature (London) 435, 922 (2005); A. Gade
et al., Phys. Rev. C 71, 051301(R) (2005); S. Grevy et al., Eur.
Phys. J. A 25, 111 (2005); J. Fridmann et al., Phys. Rev. C 74,
034313 (2006); B. Jurado et al., Phys. Lett. B 649, 43 (2007);
014302-9
PHYSICAL REVIEW C 90, 014302 (2014)
yet completed this part of the task that would make it possible
to obtain predictions for the neutron separation energies.
L. Gaudefroy et al., Phys. Rev. Lett. 97, 092501 (2006);
A. Gade et al., Phys. Rev. C 74, 034322 (2006); C. M. Campbell
et al., Phys. Rev. Lett. 97, 112501 (2006); ,Phys. Lett. B 652,
169 (2007).
[15] L. Gaudefroy et al., Phys. Rev. Lett. 102, 092501 (2009);
M. De Rydt et al., Phys. Rev. C 81, 034308 (2010);
C. Force et al., Phys. Rev. Lett. 105, 102501 (2010);
D. Santiago-Gonzalez et al., Phys. Rev. C 83, 061305(R) (2011).
[16] Y. Utsuno et al., Phys. Rev. C 86, 051301 (2012).
[17] B. A. Brown and B. H. Wildenthal, Annu. Rev. Nucl. Part. Sci.
38, 29 (1988).
[18] E. Caurier, G. Martı́nez-Pinedo, F. Nowacki, A. Poves, and
A. P. Zuker, Rev. Mod. Phys. 77, 427 (2005).
[19] A. P. Zuker, J. Retamosa, A. Poves, and E. Caurier, Phys. Rev.
C 52, R1741 (1995).
[20] A. Arima, M. Harvey, and K. Shimizu, Phys. Lett. B 30,
517 (1969); K. Hecht and A. Adler, Nucl. Phys. A 137, 129
(1969).
[21] J. P. Elliott, Proc. R. Soc. London, Ser. A 245, 128 (1956).
[22] C. E. Svensson et al., Phys. Rev. Lett. 85, 2693 (2000).
[23] E. Ideguchi et al., Phys. Rev. Lett. 87, 222501 (2001).
[24] E. Caurier, J. Menéndez, F. Nowacki, and A. Poves, Phys. Rev.
C 75, 054317 (2007).
[25] M. Dufour and A. P. Zuker, Phys. Rev. C 54, 1641 (1996).
[26] M. Seidlitz et al., Phys. Lett. B 700, 181 (2011).
[27] V. Tripathi et al., Phys. Rev. C 77, 034310 (2008).
[28] P. Doornenbal et al., Phys. Rev. Lett. 111, 212502 (2013).
[29] W. A. Richter, S. Mkhize, and B. A. Brown, Phys. Rev. C 78,
064302 (2008).
[30] J. R. Terry et al., Phys. Rev. C 77, 014316 (2008).
[31] P. Doornenbal et al., Phys. Rev. Lett. 103, 032501 (2009).
[32] T. Nakamura et al., Phys. Rev. Lett. 103, 262501 (2009);
M. Takechi et al., Phys. Lett. B 707, 357 (2012).
[33] A. Poves et al., Phys. Scr. T150, 014030 (2012).
[34] F. Rotaru et al., Phys. Rev. Lett. 109, 092503 (2012).
[35] G. Klotz et al., Phys. Rev. C 47, 2502 (1993).
[36] C. M. Mattoon et al., Phys. Rev. C 75, 017302 (2007).
[37] S. Takeuchi et al., Phys. Rev. C 79, 054319 (2009).
[38] P. Doornenbal et al., Phys. Rev. C 81, 041305(R) (2010).
[39] A. Gade et al., Phys. Rev. C 83, 044305 (2011).
[40] H. Sakurai, Astrophysics and Nuclear Structure: International
Workshop XLI on Gross Properties of Nuclei and Nuclear
Excitations, Hirschegg, Austria, 2013 (unpublished).
[41] W. Mittig et al., Eur. Phys. J. A 15, 157 (2002).
[42] S. Kahana, H. Lee, and C. Scott, Phys. Rev. 185, 1378
(1969).
[43] K. Sieja and F. Nowacki, Nucl. Phys. A 857, 9 (2011).